Feasibility study on the design, manufacturing and economic potential of fibre reinforced polymer lattice structures for wind turbine towers

Abstract

Cost is a design driver in any commercial business, and this is not different in the wind energy industry. The potential of cost savings on wind turbine towers through lightweight composite designs has been shown. However composite grid stiffened and lattice structures, which were shown to lead to significant weight reductions in aerospace applications, have not yet been under investigation for applications in wind turbine towers. In an effort to combine the advantages of composite lattice structures on structural weight with the expected cost savings through lightweight wind turbine towers, and with the observation that potential additional benefits in the airflow transparent nature of a composite lattice structure can be found, a feasibility study on a composite lattice wind turbine tower was proposed. The feasibility of this structure has been determined based on the objectives to: - Design a full scale composite lattice tower structure fulfilling the design requirements obtained from references. - Show manufacturability of representatively scaled samples of the structure. - Evaluate the real life mechanical performance and behaviour of composite lattice geometries under wind turbine typical design loads. - Obtain a relative cost figure for the composite design in comparison to a reference tower, based on the cost related to the process of tower production up to the assembly on-site. To achieve these objectives a literature review was performed on the state of the art in wind turbine design, and on the production and analysis of grid stiffened composites. A reference wind turbine, international regulations and design load references were identified as a good starting point for tower design. Literature on grid stiffened composite structures supports the expected weight saving potential of these structures. Based on the literature found an analytical smeared stiffness method was considered to be a suitable design tool to quickly obtain a preliminary design at the start of the project. Finite element modelling was identified as the design tool to perform more detailed modelling of the structure. Finite element modelling was also identified to be used in combination with a genetic algorithm to perform a weight optimisation on the tower design. Thermal expansion based tooling was identified as a manufacturing method which has effectively been used in the production of grid stiffened composite structures, but is a relatively expensive production method. Dry winding in combination with vacuum infusion at room temperature was identified as a relatively new and low cost out-of-autoclave production method and has therefore been used in this project. The reference turbine chosen to be used in this project is the "NREL 5MW reference turbine". This reference turbine is a virtual open source turbine based on properties of real life wind turbines and wind turbine design studies. Since the wind turbine data is open-source, the turbine has already been used as baseline in multiple studies. This was particularly of use as a reference in determining the design loads. The design loads used in this project were based on loads in normal wind turbine operations with a once in 50 year occurrence probability. The design loads used in this project were determined using a load probability distribution for the reference turbine during normal operations. The expected maximum loads the wind turbine tower can face in a 50 year timespan during normal operations were extracted form this distribution and used as design input. An analytical smeared stiffness optimisation method was used to obtain an expected range of composite lattice configurations. The outer bounds of this range, a large and small helical rib angle configuration, were used as input for the design of small scale glass fibre samples with which the vacuum infusion manufacturing method was determined. Through some iterations a manufacturing method was found which left no visual defects in the parts. The presence of visual defects in the parts was furthermore not found to be affected by helical rib angle. For the detailed design the final constraints considered were expanded to capture the natural frequency of the system, a minimum natural frequency constraint of 0.222 Hz has been determined based on the rotor operating frequencies. Modelling the forces generated by the weight of the tower itself was incorporated by implementing the distributed tower mass and a point mass representing the nacelle and rotor weight in combination with a gravity load. Furthermore the material modulus and strength properties required as input for FEM modelling in Abaqus were in detail determined for both glass and carbon fibre reinforced epoxies by using reference material test properties. A tri-hexagonal grid, also know as kagome grid configuration, has been chosen to make sure the node complexity was as simple as possible to reduce the risk of vacuum infusion problems. This risk reduces since only two ribs are crossing each other at nodes instead of three ribs crossing each other in one location for a triangular configuration. The geometry of the tower has been based on the reference tower but an extra base diameter constraint of 4.4m has been implemented to reduce excessive land based transportation costs for the tower sections. The composite lattice structure has been parametrized for effective optimisation based on the dimensional constraints and five rib geometry parameters: - Number of helical rib - Number of hoop ribs - Helical rib width - Hoop rib width - Rib height The ribs have been modelled using shell elements in the length and height direction of the ribs. The loads and nacelle plus rotor mass were introduced at a point in the combined centre of mass of the nacelle and rotor. This point was kinematically connected to the tower top and the tower bottom was clamped. Based on the first modelling results it was found that a glass fibre design could not at the same time be geometrically feasible and stiff enough to comply with the minimum natural frequency constraint when the tower has a base diameter of 4.4m. Using initial designs, a re-evaluation of the effect of the 4.4m base diameter constraint on transport and material cost was made using a carbon fibre design with a base diameter of 4.4m and a glass fibre design with a base diameter of 6m. It turned out that the carbon design was estimated to be 500,000 Euro less expensive. Therefore the carbon fibre alternative with a base diameter of 4.4 meter was favoured. Some model simplifications have been made in the modelling of the tower to reduce computational time. The natural frequency of the system has been accurately fitted to the tower top deflection to omit a separate frequency analysis and rib width tapering, non-linear effects and mesh convergence effects have been accounted for by means of knock-downs. A python genetic algorithm has been used to perform the optimisation. The python script automatically runs a Catia macro to update the design, and an Abaqus script to perform the analysis. The python script uses the Abaqus results on tower weight, max strain failure criterion and tower top deflection to determine the individual's fitness based on structural weight with penalty constraints on the strength factor of safety and tower top deflection. The genetic algorithm has been executed three times, with tighter ranges for the structural parameters in every step based on the results obtained in the former step. This optimisation approach led to a most optimal design found with a tower weight of 134.5 tonnes, a strength factor of safety of 2.41, and a natural frequency of 0.225 Hz. A mechanical test sample has been designed, to perform compression tests on a scaled section of the composite lattice tower. Specific attention has been given to the helical rib intersections since the helical rib width dimension led to large nodes in these locations. The load capability of the test bench was driving in the scaling performed which led to a scaling factor of 9.5 with an additional reduction of the rib height to the dimension of the helical rib width. It is a flat sample featuring three helical rib intersections. The hoop ribs on the outer edges of the samples are potted. The potting serves as a load introduction and sample stabilisation zone. Six samples of similar geometry are tested. The test data shows that the samples behave less stiff and are able to bear lower loads than the initial model predicts. The samples also show out of plane deflection behaviour which was not observed in the initial model. It has been found that the in-plane stiffness deviation was mainly due to smaller test sample rib cross-sectional areas than initially modelled, and slightly different fibre volume fractions in the test sample. And the out of plane deflection in the sample can be explained by the real life fibre distribution and waviness of the fibres in the sample. A rough progressive failure estimation has been performed using the improved model. The failure estimation was based on a maximum strain criterion and removing the element's stiffness when the element passed the maximum allowed axial strain. Using this the correct initial failure location can be estimated. The failure load found is 67 kN, where the test data shows a spread in failure load from 44 kN to 76 kN. The cost of the full scale structure has been determined based on the process of manufacturing tower sections to installing it on-site. This has been done by estimating material, production and possible savings on transport and assembly costs. The material cost has been determined based on the weight of the optimised design, the production cost has been compared with research performed on wind turbine blade production cost, and the transport and assembly savings were based on a wind turbine logistics reference study. It has been found that the material cost dominate the total cost of the composite tower, and that the composite tower is 1.7 million Euro more expensive than the steel tower reference. This cost increase boils down to an increase in cost of off-shore wind energy of 4.6-7.4\% which has to be compensated by the indicated potential additional benefits of the composite lattice tower to make the composite structure economically interesting. Based on the results of this study it has been shown that a geometrically feasible structure can be designed that complies to all design constraints. The manufacturability of small scale samples has been demonstrated, and the real life behaviour of the small samples has been tested in compression. Based on the test results the prediction of stiffness, strength and out-of-plane deflection have been improved. A cost estimation on the process of manufacturing tower sections up to installing the composite lattice tower on-site shows that the cost of off-shore wind energy increases by 4.6-7.4\% in comparison to steel towers. This has to be compensated by the indicated potential additional benefits of the composite lattice structure to make the composite structure economically interesting The final conclusion therefore is that a composite lattice tower for wind turbines shows to be structurally very promising, however further research is required into additional benefits of the structure to determine the final economic potential of the structure.